Hourly Day-ahead Solar Energy Prediction For Supporting Smart Grid Implementation in Semau Island

Knowledge of solar power prediction is an important aspect for power generation planning as well as supporting operations in grid area that apply photovoltaic (PV) generators. Such an aspect is even more critical in the case of limited availability of measurement data. This paper presents a novel model for predicting daily specific solar radiation using predetermined artificial neural networks. Daily meteorological data are used to enrich the conventional artificial neural networks technique. For simulation purposes, solar data was collected at two locations with different meteorological conditions, namely Kupang, Timor Island and Waingapu, Sumba Island, both in East Nusa Tenggara province, Indonesia. Finally, the proposed method shows good performance so as to achieve the above mentioned prediction purposes.